Short-form video overuse and adolescents’ restrained eating: the roles of body shame and physical exercise Authors: Xu Kong, Mingwei Deng, Sicheng Xiong, Chengwei Zeng, Zuhua Zhang, Chunling He, Bin Zhang Published: 2025 ABSTRACT BackgroundThe social media’s dissemination of norms for body shape is progressively shaping individual perceptions and behaviors. Social media plays a key role in causing body shame and restrained eating behavioral pattern of adolescents. Although the relationships among short-form video overuse, body shame, and restrained eating are well documented, it is unclear whether the specific problem of short-form video overuse also plays a role in these relationships. We tested body shame as a mediator and physical exercise as a moderator of the association between short-form video overuse and restrained eating among adolescents.MethodsAdolescents (N = 535; ages 13 to 17; 50.1% girls) from four secondary schools in Hunan Province, China were surveyed by cluster sampling method. The students completed the Short-Form Video Overuse Scale, the Body Shame Scale, the Dutch Eating Behavior Questionnaire, and the Physical Activity Rating Scale during regular class time.ResultsThe results of regression-based analyses indicated that short-form video overuse was positively associated with restrained eating, and body shame mediated this association. The relationship between body shame and restrained eating was stronger among adolescents who engaged in low physical exercise.ConclusionExcessive viewing of Short-form video may increase the risk of body shame and restrained eating. The results highlight the potential utility of physical exercise in interventions to reduce adolescents’ restrained eating. Introduction The current trend of “thin is beautiful” is becoming increasingly widespread [1]. The University of Chinese Academy of Social Sciences, in collaboration with the Social Sciences Academic Press, released a research report titled Research on Health Behaviors of Chinese Adolescents: Data Analysis from Surveys in 13 Provinces. The study revealed that the majority of adolescents are not entirely satisfied with their body weight, and 9.2% of them opt for dietary restriction as a means of weight loss [2]. Restrained eating has become a hot topic in people’s daily lives. Restrained eating refers to a self-imposed limit on eating for a long time in order to lose weight or maintain the same weight [3]. However, restrained eating is not effective as a means of weight loss, and in fact has been shown to be associated with overweight and obesity among adolescents [4]. Restrained eating is also associated with an increased risk of eating disorders, depression, and anxiety [5, 6]. These highlight the severity of such behaviors, especially as social media and cultural norms around the “ideal body type” encourage unhealthy eating patterns among teenagers, such as extreme dieting or intermittent fasting. These trends underscore the need for greater awareness and intervention efforts to address the root causes of disordered eating and prevent associated health risks like eating disorders, depression, and anxiety. Knowledge of the predictors of adolescents’ restrained eating could inform prevention and intervention efforts to reduce this behavioral pattern. The relationship between short-form video overuse and restrained eating Internet and social media use has been shown to be associated with young women’s symptoms of eating disorders [7], negative induced emotion with their bodies [8] and eagerness to lose weight [9]. Short-form video is a novel form of online media that might also be associated with problematic eating [10]. TikTok, Instagram, and Kuaishou are well known platforms for these videos, which are posted by other users and typically last less than 90 s. These short videos are thought to cater to individuals’ needs for leisure, entertainment, social interaction, and information acquisition [11]. China has a large number of short-form video users. As of June of 2025, the number of short-form video users in China reached 1.08 billion, accounting for 96.7% of all netizens [12]. Short-form video overuse has been called a behavioral addiction because the viewer loses control over the behavior and experiences negative consequences due to excessive use [13]. Numerous studies have investigated the relationship between short-form video usage and restrained eating, with a particular emphasis on how idealized body images portrayed in media can affect individuals, mainly involving objectification, body image comparison and the emotions that arise from it [8, 14–16]. Research indicates that frequent engagement with short-form video, such as TikTok and Instagram, can exacerbate negative induced emotion, which in turn increases the propensity for individuals to adopt restrictive eating as a weight management strategy [17]. This suggests that the portrayal of unattainable beauty standards within these platforms significantly contributes to unhealthy eating behaviors aimed at achieving these idealized body images. Vygotsky’s sociocultural theory regards social culture as a key factor for individuals’ cognition in society. Social culture is composed of countless individual identities, but it also affects each individual identity [18]. For some adolescents, short-form video social media platforms are a key source of information and an important channel for social and cultural communication. This insular nature of this social culture creates an “information cocoon” [19]. In fact, such “information cocoon” stem from platform algorithms. Existing research has demonstrated that social media platforms employ algorithms designed to deliver personalized content, which enhances user engagement and prolongs using time [20], thereby amplifying potential negative effects [21]. Compared to general social media, short-form video platforms employ more aggressive and excessive algorithmic models and exercise weaker platform oversight [21]—a difference attributable primarily to their distinctive “infinite scroll” interface design [22]. This design delivers a ceaseless flow of content, leading to greater loss of control over usage behavior [22]. Furthermore, studies have confirmed that exposure to ideal image in short-form videos, as opposed to static images, elicits stronger negative emotions and promotes higher levels of self-objectification among individuals [15].As a highly visual form of social media, short-form video platforms allow for the enhancement and modification of dynamic visual content tailored to user preferences, thereby presenting more idealized body images and exerting a more substantial negative impact on individuals’ body image perception [23]. Through persistent reinforcement and deepening comprehension, the “media’s ideal image” for individuals has evolved into a universally accepted standard [24]. An empirical study lends support to this perspective. After exposing 120 young women to short videos depicting idealized body types, results indicated that the exposure group reported significantly higher levels of appearance comparison and negative emotions compared to the control group [8]. They may resort to restrained eating in an attempt to alleviate anxiety and feelings of guilt regarding their physical appearance [25]. In fact, another study by Griffiths et al. analyzed 1.03 million TikTok videos that were distributed to 42 eating disorder patients and 49 healthy controls over a one-month period. The findings indicate that TikTok’s algorithm may exacerbate eating disorder symptoms through a numbing content personalization process based on users’ voluntary behaviors [26]. The mediating role of body shame According to sociocultural theory, body image-related social pressures play a significant role in the development of body shame. The Tripartite Influence Model further suggests that increased pressures regarding appearance from social media, parents, and peers contribute to body image concerns, both through mechanisms such as social comparison and the internalization of body ideals [27]. Social comparison theory maintains that individuals usually judge whether they meet standards by comparing themselves with others, and the evaluation that they do not meet the standards may lead to negative emotions [28].One possibility of negative emotions is that body shame, which refers to the negative emotion experienced by an individual who believes that their appearance does not meet the ideal standard or is negatively evaluated by others [29]. People make social comparisons while watching short-form videos, including comparisons between their own body image and the bodies of idealized figures depicted in such videos [30]. The opportunities to make these comparisons increase because social media platforms send new videos based on videos that were watched in the past; the viewer repeatedly sees similar content because of this personalized information delivery [31]. Indeed, in one study, participants reported feeling ashamed of their appearance when they perceived a gap between their own image and their perfect image [32]. We expected body shame acts as a mediator and are interested in the mediation process by which short-form video viewing increases body shame, and body shame in turn increases restrained eating. The second link in this mediation process is between the negative emotion of body shame and the behavioral pattern of restrained eating. One assumption in the literature is that people engage in restrained eating in an attempt to eliminate body shame and other negative emotions [33]. According to objectification theory, the individual’s body is socially constructed as an object observable to others within a cultural context [29, 34]. This process leads individuals to internalize this observer’s perspective, ultimately perceiving themselves as objects to be looked at and evaluated. Exposure to the “skinny beauty” trend in short-form videos for a long time can lead adolescents to internalize this standardized aesthetic and trigger negative evaluations of real body images [35]. They may feel ashamed of their less-than-ideal physical appearance and thus resort to controlling their image by restricting food intake [36]. Individuals tend to engage in maladaptive behaviors such as restricting food intake or excessive exercise to conform to the contemporary aesthetic standard [33]. Accordingly, we pose Hypothesis 2: Body shame will mediate the relationship between short-form video overuse and restrained eating. The moderating effect of physical exercise With the development of sports psychology, the role of physical exercise in improving emotional state and reducing the occurrence and development of mental health problems has been confirmed [37]. In accordance with the hierarchical hypothesis of self-concept, physical exercise contributes to the development of a positive physical self-concept and physical self-esteem [38], a conclusion that has been well established in existing research. Some studies have demonstrated that engagement in physical activity significantly enhances individuals’ physical self-concept [39], and this enhancement further promotes adolescents’ sense of meaning in life through the mediating role of self-concept [40]. Notably, studies by Gilchrist, Crocker, and others have indicated that lower levels of physical self-concept are often accompanied by higher levels of body shame and guilt [15, 41]. Based on this, we infer that body shame—an essential negative emotion closely tied to body image—may be mitigated through regular physical exercise. This is particularly relevant in the current context of widespread short-video culture, where intensified social comparison pressures heighten individuals’ susceptibility to body shame [32]. Thus, physical exercise may function as a protective factor that weakens the association between excessive short-video use and body shame, thereby moderating the initial stage of this mediating pathway. Furthermore, physical exercise may also moderating the second stage of the mediating pathway—namely, the relationship between body shame and dieting behavior. Early research has shown that individuals who adopt restrictive eating behaviors often exhibit an unstable self-concept and low self-esteem [42]. Other studies have found that when individuals objectify themselves and excessively focus on appearance, they are more prone to restrictive dieting, which may directly or indirectly trigger binge eating [43].) These findings suggest a close link between self-concept and eating behavior. Given that body shame has been confirmed to correlate strongly with low self-esteem, and physical exercise can effectively enhance physical self-esteem and improve perceptions of bodily worth, physical ability, and physical attractiveness, we further hypothesize that if physical exercise strengthens physical self-esteem and self-concept, it may also reduce body shame, thereby weakening the association between body shame and dieting behavior. Empirical evidence supports this assumption. Shang’s research indicates that physical exercise helps participants develop a more positive physical image, which further promotes the subjective well-being of college students [44]. Te Velde et al. likewise found a positive relationship between psychological self-concept and physical activity, with active individuals demonstrating more positive self-evaluations than those who do not exercise [45]. A 6-month physical exercise program (running or mixed activities) was administered to 57 participants. Measures of self-concept and well-being were assessed at baseline, post-intervention, and at a 6-month follow-up. Results demonstrated that compared to the control group, the exercise group showed significant improvements in physical self-concept and reduced psychosomatic symptoms [39]. These improvements not only help alleviate negative emotions but also reduce the risk of disordered eating [35, 46]. Thus, we pose two hypotheses: Hypothesis 3: Physical exercise will attenuate the relationship between short-form video overuse and body shame. Hypothesis 4: Physical exercise will attenuate the relation between body shame and restrained eating. Hypothesis 3: Physical exercise will attenuate the relationship between short-form video overuse and body shame. Hypothesis 3: Physical exercise will attenuate the relationship between short-form video overuse and body shame. Hypothesis 4: Physical exercise will attenuate the relation between body shame and restrained eating. Hypothesis 4: Physical exercise will attenuate the relation between body shame and restrained eating. The current study To address the above-mentioned research gaps in the literature, we tested the association between adolescents’ overuse of short-form videos on social media and their restrained eating. We also tested the mediating role of body shame and moderating role of physical exercise in this association. This knowledge can inform parents, teachers, and educational professionals for the prevention and intervention of adolescent restrained eating (Fig. 1). Fig. 1The Hypothesized Model The Hypothesized Model Methods Participants The sample consisted of 535 adolescents (50.1% were girls). The cluster random sample was formed in four secondary schools in Changsha, Hunan Province, China. In the selected schools, all 8th and 9th grade students were invited to participate. The mean age of the study sample was 14.21 (SD = 0.66) years, and the age range was from 13 to 17 years. Procedure This study was part of a larger interdisciplinary project examining social media use and restrained eating tendencies among adolescents. The study was conducted with the approval of the school directors and with parental consent. Students completed self-report questionnaires in their classrooms. A total of 556 students were invited to participate. Of these, 21 questionnaires were excluded because more than 20% of items were left unanswered or because patterned or uniform responding was detected. For the final analysis, 535 valid questionnaires were retained. For the retained data, the amount of missing values was minimal; therefore, item-level mean imputation was applied to handle the missing data. Measures Short-Form video overuse scale Short-form video overuse was assessed using the self-developed Short-Form Video Overuse Scale, which was adapted from the Facebook Intrusion Questionnaire [47]. The original measure assessed the person’s preoccupation with Facebook, a social media platform. First, the Facebook Intrusion Questionnaire was translated into a Chinese version and validated. Then the word “Facebook” was replaced with “short-form video platform.” The questionnaire consists of 8 items, such as “When I’m not using the short-form video platform, I often think about it.” Participants rated each item on a five-point scale (1 = completely disagree, 5 = completely agree). Higher scores indicate more short-form video overuse [48]. In this study, the Cronbach’s α was 0.91. The results of the CFA indicated an acceptable model fit for our measurement model: χ²/df = 5.00, RMSEA = 0.08, TLI = 0.90, CFI = 0.93, SRMR = 0.04. Body shame scale Body shame was measured using the Chinese version of the Body Shame scale compiled by Durate et al. and revised by Sun Wan et al. [49]. The scale consists of 12 items, such as “Wearing clothes that show my body shape easily makes me feel bad.” Each item was rated using a 6-point scale (1 = never, 6 = always). A higher total score indicates more body shame. Cronbach’s α in this study was 0.92. Restrained eating scale Restrained eating was measured using the Restrained Eating Subscale from the Dutch Eating Behavior Questionnaire [50]. The scale consists of 10 items, each rated using a 5-point scale (1 = never, 5 = always), with higher scores indicating more restrained eating. In this study, Cronbach’s α for this scale was 0.94. Physical activity rating scale Physical exercise was measured using the three-item Physical Activity Rating Scale [51]. Respondents rate their exercise on the dimensions of intensity, frequency, and duration. Intensity and frequency are rated on a scale of 1 to 5, and duration is scored on a scale from 0 to 4. The exercise score is calculated by multiplying the ratings for intensity, duration, and frequency, with possible scores ranging from 0 to 100. Scores of 19 and below indicate low exercise; scores between 20 and 42 indicate moderate exercise; scores of 43 and above indicate high exercise. In this study, the test-retest reliability of the scale was 0.85. Data analysis The statistical analyses were performed using SPSS 26.0. After descriptive statistics and Pearson correlations were generated, the proposed mediation model and moderated mediation model were tested using the SPSS macro PROCESS version 4.0 (Model 4 and Model 58, respectively) developed by Hayes. Results Test of common method bias Harman’s single factor test was used to investigate common method bias. The variation in the first factor explained 30.13% of the total variance. This value was less than the cutoff of 40% [52], indicating that common method bias was not a substantial problem. Descriptive statistics Descriptive statistics and correlations were computed for all variables and are presented in Table 1. The correlations were significant and in the expected directions. All bivariate correlations ranged from − 0.22 to 0.38, well below the commonly accepted cutoff of 0.70, indicating no concerns regarding multicollinearity. Consistently, all variance inflation factor (VIF) values were 1.00. As Gender was significantly related to the main research variables, it was included as a covariate in the subsequent analyses. Table 1Descriptive statistics and correlations among variablesVariableMSD123451.Gender1.50.5012. Short-form video overuse2.060.870.10*13. Body shame2.160.850.20**0.32**14. Restrained eating2.191.000.10*0.16**0.38**15. Physical exercise1.730.76−0.21**−0.11**−0.22**−0.021N = 535. *p < 0.05. **p < 0.01. Male = 0, Female = 1 Table 1Descriptive statistics and correlations among variablesVariableMSD123451.Gender1.50.5012. Short-form video overuse2.060.870.10*13. Body shame2.160.850.20**0.32**14. Restrained eating2.191.000.10*0.16**0.38**15. Physical exercise1.730.76−0.21**−0.11**−0.22**−0.021N = 535. *p < 0.05. **p < 0.01. Male = 0, Female = 1 Descriptive statistics and correlations among variables N = 535. *p < 0.05. **p < 0.01. Male = 0, Female = 1 Test of the mediation model Model 4 from the SPSS macro PROCESS was used to test the mediating effect of body shame on the relationship between short-form video overuse and restrained eating. The results (Table 2) showed that after controlling for gender, a significant positive correlation was found between short-form video overuse and restrained eating. (βs = 0.18, t = 3.59, p = 0.0004). The direct effect was no longer significant (βs = 0.05, t = 1.06, p = 0.29) when body shame was added as a mediator. There was a significant positive correlation between short-form video overuse and restrained eating. (βs = 0.30, t = 7.44, p < 0.001), and body shame showed a positive correlation with restrained eating (βs = 0.42, t = 8.38, p < 0.001). The percentile bootstrap method with deviation correction revealed that body shame partially mediated the relationship between short-form video overuse and restrained eating, with an estimated indirect effect (ab) of 0.13, standard error (SE) of 0.02, and a 95% confidence interval of [0.08, 0.18]. The mediation effect explained 70.62% of the total effect.Table 2Summary table of mediation effect analysisPredictorBody shameRestrained eatingΒt95% CIΒt95% CIGender0.284.08***[0.15, 0.42]0.050.55[−0.12, 0.21]short-form video overuse0.307.44***[0.22, 0.38]0.051.06[−0.04, 0.15]Body shame0.428.38***[0.32, 0.52]R 20.130.15F39.49***30.28***N = 535. Bootstrap sample size = 5000. CI = confidence interval. Β = standardized coefficient. *p < 0.05. **p < 0.01. Male = 0, Female = 1 Table 2Summary table of mediation effect analysisPredictorBody shameRestrained eatingΒt95% CIΒt95% CIGender0.284.08***[0.15, 0.42]0.050.55[−0.12, 0.21]short-form video overuse0.307.44***[0.22, 0.38]0.051.06[−0.04, 0.15]Body shame0.428.38***[0.32, 0.52]R 20.130.15F39.49***30.28***N = 535. Bootstrap sample size = 5000. CI = confidence interval. Β = standardized coefficient. *p < 0.05. **p < 0.01. Male = 0, Female = 1 Summary table of mediation effect analysis N = 535. Bootstrap sample size = 5000. CI = confidence interval. Β = standardized coefficient. *p < 0.05. **p < 0.01. Male = 0, Female = 1 Test of the moderated mediation model Model 58 from the SPSS macro PROCESS was used to test the proposed moderated mediation model. Moderation was tested in both parts of the mediation process: that is, physical exercise was expected to mitigate the association between short-form video overuse and body shame, and to mitigate the association between body shame and restrained eating. The results (Table 3) showed that contrary to Hypothesis 3, the interaction between short-form video use and physical exercise did not show a positive correlation with body shame (βs = − 0.02, t = −0.59, p = 0.55). However, the interaction between body shame and physical exercise significantly, negatively correlated restrained eating (βs = − 0.09, t = −2.18, p = 0.03) providing support for Hypothesis 4. The full mediated moderation model is depicted in Fig. 2. The model predicting the mediator, body shame, explained 15% of the variance (R² = 0.15, F(4, 530) = 24.05, p < 0.001). The model predicting the outcome, restrained eating, accounted for 16% of the variance (R² = 0.16, F(5, 529) = 19.93, p < 0.001). The inclusion of the body shame × physical exercise interaction term resulted in a ΔR² = 0.008, indicating a modest increase in the model’s explanatory power due to the moderation effect. For the conditional indirect effects of short-form video overuse on restrained eating through body shame, the indirect effect was significant at different levels of physical exercise. Specifically, the indirect effect was 0.14 when physical exercise was one standard deviation below the mean, 0.10 at the mean, and 0.07 when physical exercise was one standard deviation above the mean. These effects were significant at all levels of the moderator, with the confidence intervals for the indirect effects not containing zero (See Table 4). Fig. 2Standardized Path Coefficients of the Hypothesized Model Standardized Path Coefficients of the Hypothesized Model Table 3Summary of moderated mediation effect analysisPredictor variableBody shameRestrained eatingΒt95%CIΒt95%CIGender0.273.24**[0.10, 0.43]0.080.96[−0.08, 0.24]Short-form video overuse0.297.13***[0.21, 0.37]0.051.10[−0.04, 0.13]Physical exercise−0.16−3.87***[−0.24, −0.08]0.061.40[−0.02, 0.14]Body shame0.368.20***[0.27, 0.45]Short-form video overuse × Physical exercise−0.02−0.59[−0.10, 0.05]Body shame × Physical exercise−0.09−2.18*[−0.18, −0.01] R² 0.150.16 F 24.05***19.93***N = 535. Bootstrap sample size = 5000. CI = confidence interval. Β = standardized coefficient. *p < 0.05. **p < 0.01. Male = 0, Female = 1 Table 3Summary of moderated mediation effect analysisPredictor variableBody shameRestrained eatingΒt95%CIΒt95%CIGender0.273.24**[0.10, 0.43]0.080.96[−0.08, 0.24]Short-form video overuse0.297.13***[0.21, 0.37]0.051.10[−0.04, 0.13]Physical exercise−0.16−3.87***[−0.24, −0.08]0.061.40[−0.02, 0.14]Body shame0.368.20***[0.27, 0.45]Short-form video overuse × Physical exercise−0.02−0.59[−0.10, 0.05]Body shame × Physical exercise−0.09−2.18*[−0.18, −0.01] R² 0.150.16 F 24.05***19.93***N = 535. Bootstrap sample size = 5000. CI = confidence interval. Β = standardized coefficient. *p < 0.05. **p < 0.01. Male = 0, Female = 1 Summary of moderated mediation effect analysis N = 535. Bootstrap sample size = 5000. CI = confidence interval. Β = standardized coefficient. *p < 0.05. **p < 0.01. Male = 0, Female = 1 Table 4Mediating effect of body shame between Short-form video overuse and restrained eating at different physical exercise levelsPhysical exercise levelMediating effect valueBoot standard error95%CILower limitUpper limit M-SD 0.140.030.080.20 M 0.100.020.070.14 M + SD 0.070.020.030.12 Table 4Mediating effect of body shame between Short-form video overuse and restrained eating at different physical exercise levelsPhysical exercise levelMediating effect valueBoot standard error95%CILower limitUpper limit M-SD 0.140.030.080.20 M 0.100.020.070.14 M + SD 0.070.020.030.12 Mediating effect of body shame between Short-form video overuse and restrained eating at different physical exercise levels Simple slopes analysis was used to interpret the significant interaction between body shame and physical exercise as a predictor of restrained eating. Figure 3 shows the association between body shame and restrained eating at low (M − 1SD) and high (M + 1SD) levels of physical exercise. In the low physical exercise group, body shame was significantly associated with restrained eating (simple slope = 0.45, t = 8.08, p < 0.001). In the high physical exercise group, body shame was also significantly associated with restrained eating (simple slope = 0.27, t = 4.07, p < 0.001), but the effect was smaller than in the low physical exercise group. The smaller effect size suggests that high physical exercise buffered the relationship between body shame and restrained eating. Fig. 3Simple Effect Analysis of Body Shame and Physical Exercise Simple Effect Analysis of Body Shame and Physical Exercise Discussion The present study examined the association between short-form video overuse and restrained eating, as well as a mediating and a moderating effect involved in this association. Results showed that short-form video overuse was positively associated with restrained eating, and this association was mediated by body shame. Additionally, physical exercise weakened the relationship between body shame and restrained eating. The findings not only contribute to an understanding of how short-form video overuse might lead to adolescents’ disordered eating, but also provide a broad perspective that can inform prevention and intervention efforts to reduce disordered eating disorders in this population. The relationship between short-form video overuse and restrained eating Short-form video overuse was significantly positively correlated with restrained eating, aligning with previous research [6]. Short-form video platforms are often inundated with various idealized body images and lifestyles, leading individuals to feel dissatisfied with their own body image and attempt to adjust their appearance through restrained eating to conform to societal beauty standards [10]. The unique algorithmic mechanisms and design features of short-form video platforms are likely to exacerbate these psychological pathways. Furthermore, adolescence is an important period for the formation of self-identity and body image, during which individuals are often more susceptible to media influence [53]. Celebrities and internet influencers may talk about behaviors associated with restrained eating, such as limiting food intake, adhering to specific dietary patterns, or emphasizing weight control [54]. Such behaviors can be widely disseminated and imitated, particularly among adolescents. Therefore, adolescents who overuse short-form videos may imitate these behaviors. The mediating role of body shame The direct effect of short-form video overuse on restrained eating was not significant, but indirectly through the mediation of body shame. This result further confirmed that body shame is an important mechanism by which short-form video overuse might trigger non-adaptive eating behavior [25]. The constant promotion of “ideal beauty” in short-form videos may accelerate adolescents’ internalization of ideals. Adolescents may engage in upward social comparisons to the people they see online and find themselves inadequate [23]; this perception in turn may trigger anxiety, shame, and other negative emotions. To alleviate this shame, adolescents may restrict their eating to lose weight [33]. The moderating role of physical exercise We found that physical exercise did not moderate the first path of the mediation pathway. That is, contrary to expectations, exercise did not weaken the association between short-form video overuse and adolescents’ body shame. Self-Determination Theory offers a valuable lens, physical exercise motivation can be divided into internal motivation and external motivation [55]. Behavior that is internally motivated is driven by internal forces, reflecting an individual’s natural propensity for learning and growth. Behavior that is externally motivated is driven by external forces, treating physical exercise as a means to attain positive outcomes or avoid negative emotions. As students progress in grade and age, academic workload increases, leading to more sedentary lifestyles, which may also be one of the contributing factors. Additionally, the use of devices like mobile phones has increased screen time for students [56]. Overall, adolescents show a lack of awareness about physical exercise; they also appear to lack intrinsic motivation and engage in limited autonomous exercise [57], physical exercise often occurs after individuals become aware of the discrepancy between their actual body image and their ideal body image. Although physical exercise can help counteract negative emotions, most adolescents engage in it based on external motivations such as weight loss or body shaping. Consequently, its protective effect tends to occur after the onset of body shame, rather than preventing it in advance. In contrast to the lack of moderation in the first part of the mediation process, physical exercise did moderate the second path of the mediation process. Physical exercise attenuated the relation between body shame and restrained eating. Although the association between body shame and restrained eating was statistically significant in both the low-exercise group and the high-exercise group, but the effect was weaker in the latter group.This finding lends support to the hierarchical model of self-concept, which posits that engagement in physical activity can enhance an individual’s physical self-concept and body esteem [38]. Positive physical self-concept and body-esteem in adolescents entail a heightened focus on the health and functionality of the body, that may emphasize a holistic evaluation of their bodies beyond mere outward appearance. The another reason could be that physical exercise can effectively improve the individual’s emotional state [37]. Positive emotions might counteract negative emotions like body shame, thus reducing the likelihood of resorting to restrained eating as a means to alleviate body shame [58]. Indeed, a positive self-concept also helps foster more positive emotions in individuals. In essence, this perspective aligns with the previous viewpoint and further supports our finding. Therefore, engaging in high physical exercise could lessen negative emotions like body shame, reducing the occurrence of restrictive eating behaviors. Implications and limitations The results of this study have two potential applications for prevention and intervention. On the one hand, adolescents should be guided to develop a diverse understanding, internalization, and pursuit of the meaning and essence of “beauty.” Engagement in school activities related to physical exercise can promote positive self-esteem [59], thereby reducing unreasonable self-scrutiny and shame. On the other hand, adolescents should be taught media literacy so that they are more critical consumers of social media’s presentation of body ideals. At the same time, this study had the following limitations. First, many factors, such as family environment and personality traits, may influence body shame and restrained eating [60, 61]. Testing more complex models will allow us to gain deeper insights into how these factors interact with social media and are associated with restrained eating behavioral pattern in adolescents. Second, the study is limited by cross-sectional data. Intervention studies and longitudinal studies could further explore the mechanisms of the occurrence and development of restrained eating. Third, because this study relied on self-reported data from adolescents, the influence of social desirability cannot be ruled out. Future studies could use multiple sources of reporting or expanding data collection channels to enhance the validity of the results. Moreover, the findings of this study should be interpreted within the unique sociocultural context of China. All participants were drawn from Chinese society and thus shaped by its cultural influences. Unlike the “thin ideal” commonly discussed in Western contexts, the “ideal body” encountered by Chinese adolescents on short-video platforms often reflects a hybrid of local aesthetic preferences and globalized imagery. Finally, we used gender as a covariate because there were mean differences between boys and girls on some of the study variables, but we did not delve further into those differences. This could be the focus of future research. Conclusion Short-form video overuse was significantly positively correlated with restrained eating. Body shame mediated the relationship between short-form video overuse and restrained eating. Moreover, physical exercise moderated the second path of the mediation processes. Compared with students who engaged in less physical exercise, those who engaged in more physical exercise demonstrated a weaker association between body shame and restrained eating.